AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 146:197–208 (2011)

Status and Health in Roman : The Effect of Status on Risk of Mortality in Post-Conquest Populations Rebecca C. Redfern1 and Sharon N. DeWitte2*

1Department of Archaeological Collections and Archive, Centre for Human Bioarchaeology, Museum of , London EC2Y 5HN, UK 2Department of Anthropology, University at Albany, Albany, NY 12222

KEY WORDS ; mortality; sex differences

ABSTRACT The Roman conquest of Britain was pre- Gompertz–Makeham models of mortality. The results of viously shown to have negatively impacted health, par- these analyses indicate that high-status individuals, par- ticularly for children, older adults, and men. We build ticularly children, had a lower mortality risk compared upon this previous research by investigating the effect to lower-status groups; and for those buried in urban that status had on risks of mortality within the Roman cemeteries, higher-status individuals of all age-groups Britain populations of Dorset. This study incorporates a had a lower mortality risk. As with our previous study sample of 291 individuals excavated from several ceme- (Redfern and DeWitte: Am J Phys Anthropol 144 (2011) teries in the county of Dorset dating between the first to 269–285), we found that male mortality risk was higher early fifth centuries AD. To assess the effect of status on than females, which we consider to reflect underlying risks of mortality, burial type was used as a proxy for sex-differences in immunity and disease response. Am J status and modeled as a covariate affecting the Siler and Phys Anthropol 146:197–208, 2011. VC 2011 Wiley-Liss, Inc.

The Roman conquest of Britain by the Emperor Clau- in the ancient world, as evidenced in medical texts (e.g., dius in 43 AD resulted in considerable changes to the Hippocrates—airs, waters, and places). Society and cul- landscape, economy, and indigenous culture and incorpo- ture can influence a person’s health, because they per- rated Britain into an Empire, which at the height of its meate and often determine the entire life course (Rou- powers (third century BC to fifth century AD), united sham and Humphrey, 2002). People who are disadvan- communities in the Arab World, North Africa, and taged socially and economically have a higher risk of Europe (Wells, 1992). This process is known as ‘‘Romani- developing a serious illness and have a shorter life ex- zation,’’ a term which has recently undergone significant pectancy (Nettle, 2010). The disadvantaged have higher changes to move away from its imperialist and colonial morbidity and mortality risk because of the cumulative past (e.g., Hingley, 2005; Gosden, 2006) and, instead, is effects of inadequate diet and poor living conditions. It is now used as ‘‘a convenient denomination covering the now well established that health status during childhood events involved in the creation of Roman [Britain], with directly affects an individual’s long-term health out- no cultural implications taken for granted’’ (Terrenato, comes, such as the development of metabolic and infec- 1998: 20). Work by the authors (Redfern and DeWitte, tious diseases because of living conditions and poor diet 2011) has demonstrated that this process had significant (McElroy and Townsend, 1996; Wilkinson and Marmot, biological repercussions, because it negatively impacted 2003). health in all age-groups and both sexes, particularly for To further investigate the results of our previous subadults and adult males. This finding was unexpected, study, we used the Siler and Gompertz–Makeham mod- as the Roman world was male orientated (Garnsey, 1968; els of mortality to investigate the relationship between Rawson, 1986b; Wiedemann, 1989). Our research sug- age, sex, and status by testing the hypothesis that gests that underlying biological differences between the higher status, as evidenced by the type of coffin used for sexes allowed females to better buffer risk, probably burial, was associated with lower mortality risk in through immunological advantages, despite them being Roman Britain. By examining how risk of mortality perceived as having a lower status than males in this pe- riod (Stinson, 1985; Flemming, 2000; Langley, 2003). We believe that these risks were created by the introduction Grant sponsors: University of Birmingham and Dorset Natural of urban settlements and increased population migration History and Archaeological Society, University at Albany Center for and the adoption of ‘‘Romanized’’ life-styles, particularly Social and Demographic Analysis (CSDA), and the University at Albany Research Foundation. how children were cared for [see also Gowland and Red- fern (2010)]. The research presented here expands upon *Correspondence to: Sharon DeWitte, Anthropology, University at our previous findings by focusing on the relationship Albany, AS 237, 1400 Washington Avenue, Albany, NY 12222. between social status, age, and sex, as these factors were E-mail: [email protected] highly influential in the Roman world (Harlow and Lau- rence, 2002). Received 2 December 2010; accepted 27 April 2011 The important relationship between health and social status has been explored by numerous bioarchaeological DOI 10.1002/ajpa.21563 studies at the population scale of analysis (e.g., Steckel Published online 8 August 2011 in Wiley Online Library and Rose, 2002; Cardoso, 2007) and was also recognized (wileyonlinelibrary.com).

VC 2011 WILEY-LISS, INC. 198 R.C. REDFERN AND S.N. DEWITTE varied with social status, we hope to reveal a factor that servicing a public fountain and a forum. Excavations contributed to heretofore ‘‘hidden’’ heterogeneity in have revealed evidence for industrial activities, such as frailty (Vaupel and Yashin, 1985; Wood et al., 1992) in metal working, and it is suggested that the environs the population of Roman Britain. were managed (Woodward et al., 1993). Outside the town, the extent to which settlements were ‘‘Romanized’’ did not depend on their distance to the principle center, SOCIAL STATUS IN THE ROMAN WORLD but, instead, it was more complex, involving a broad An individual’s status in the Roman world was negoti- range of variables (Putnam, 2007). For example, in rural ated according to their biological sex, age, and socioeco- areas, villa settlements were the epitome of Roman liv- nomic status. During this period, Roman society viewed ing as they were high-status settlements that incorpo- the adult male body as the ideal physiology, above those rated luxury styles of living (Putnam, 2007). Other rural of women and children who were regarded as being settlements were more comparable to their preceding incomplete and, therefore, of less value (Foxhall and late Iron Age counterparts, consisting of wattle and daub Salmon, 1998; Flemming, 2000). The emphasis on adult- roundhouses (Hingley, 1989). hood is crucial, as children were seen as incomplete beings that were transformed through the life course SOCIAL AND FUNERARY STATUS IN THE ROMAN into adult beings by various social and religious rituals PERIOD (Rawson, 1986b, 1991). These social notions were rein- forced by legal proscriptions, which dictated ages of In archaeology, an individual’s status and identity are responsibility in terms of property ownership and voting most frequently investigated using available funerary evi- rights (Garnsey, 1968; Harlow and Laurence, 2002). dence (e.g., Buzon and Judd, 2008). Hope’s (2009) review Recent research taking a gender or life course approach of Roman death shows that the relationship between has demonstrated that these social proscriptions were social and funerary statuses can be reliably established, subject to regional and temporal variation within the and we are fortunate to have a wide range of evidence Empire, reflecting local traditions, and show that a uni- relating to funerals and cemeteries from the Roman pe- form way of life did not exist (Gowland and Redfern, riod (Toynbee, 1996). Written evidence, including prices 2010; Revell, 2005; Carroll, 2006). Within the Empire, a and amounts to be set aside for burial, can provide impor- wide variety of status groups existed from slaves to rul- tant information about status and burial practice ing elites, and it was possible over an individual’s life (Toynbee, 1996). However, these sources are not without course to move between groups because of factors such bias in terms of geographic and temporal distribution in as forced slavery and manumission, marriage, and the Empire, and we should question how representative migration (Adkins and Adkins, 1998). they are of the community from which they derive, as we In Roman Britain, a range of status groups has been do not know whether complex high-status funerals identified using a wide range of archaeological evidence, described in texts from Rome occurred elsewhere in the including funerary and epigraphic data. At the most fun- Empire (Hope, 2009). Evidence from legal texts shows damental level, the greatest distinction was between the that people often bought land in the suburbs and built enslaved and the free. Slaves lived in both urban and ru- tombs during their lifetime; funeral clubs were also popu- ral settlements, and these individuals could have been lar in the Empire and were joined by slaves, freedmen, transported from all areas of the Empire and also from and the less prosperous to ensure that their bodies were marginal territories (Scheidel, 1997; Joshel, 2010). Other buried. The lowest status and cheapest burials just placed social and status groups in Britain included the military, the body in the ground, while more expensive coffins merchants and traders, skilled craftspeople, unskilled were made of wood, terracotta, stone, and lead (Toynbee, and agricultural workers, and government officials 1996; Hope, 2009). However, variations within each type (Birley, 1964, 1979). The population of Roman Britain existed, and, as Esmonde Cleary (1987) highlights, a has been shown by stable isotope and ancestry studies to stone coffin at source may have cost as much as a quality be very diverse, with people from the Mediterranean, wooden one elsewhere. The wealthy also used decorated other areas of Europe, and Africa (Evans et al., 2006; or plain sarcophagi made of marble, stone, or lead to Leach et al., 2010). In Dorset, the population appears to ‘‘house’’ a coffin. Those who could not afford to save (e.g., have been formed of locals and migrants from the North- slaves, the poor, and noncitizens) received basic disposal, ern Mediterranean—populations that show very little in mass-burial pits, from the local authorities or none at genetic differentiation (Richards et al., 1998; Lao et al., all and were left exposed to the elements and scavenging 2008; Redfern et al., 2010). The majority are believed to animals (Toynbee, 1996; Hope, 2009). have participated in agricultural activities, but the mate- rial culture evidence shows that the town of Dorchester BURIAL PRACTICES IN ROMAN BRITAIN (Durnovaria) must have been inhabited by skilled crafts- people, merchants, and traders; some military personnel One of the most significant impacts of the Roman appear to have lived in the region, although a perma- invasion of Britain in 43 AD was the introduction of nent fort has not been found (Woodward et al., 1993; organized urban settlements, which created a new physi- Putnam, 2007). The region was governed by an elite cal and conceptual geography for the inhabitants formed of native and newcomers, who were responsible (Esmonde Cleary, 1987; Mattingly, 2006). Burial of the for keeping order and administrating the province. dead played an influential role in this new landscape, as The walled urban settlement of Durnovaria was built the towns were surrounded by cemeteries, with graves in the Roman style, with rectangular buildings made of visible on either side of the roads linking the town to stone, wood, or a mixture of the two, with homes often other settlements (Esmonde Cleary, 1985, 2000). Ceme- having plumbing and latrines, under-floor heating (hypo- teries in Roman Britain are more common in towns com- caust), and private bathhouses (Putnam, 2007). The pared to settlements in the countryside. The majority town also had a large public bathhouse, an aqueduct dates from the third to early fifth centuries, and because

American Journal of Physical Anthropology STATUS AND HEALTH IN ROMAN DORSET 199

TABLE 1. Sample used in the study presented by site, location and numbers of individuals Ambiguous Undetermined References Site name Location Male Female sex sex Subadult Total Woodward et al., 1993 Albert Road Urban 6 1 – 1 10 18 Davies et al., 2002 Alington Avenue Rural 22 13 5 – 11 51 Smith, 1993 County Hall Urban – – – – 2 2 Green et al., 1981 Crown Building Urban 1 – – – – 1 Smith et al., 1997 Dorchester Bypass Rural 11 8 1 – 13 33 Startin, 1982 Fordington Old Vicarage Urban 5 1 – – 6 12 Smith, 1993 Greyhound Yard Urban – – – – 8 8 Wainwright, 1979 Gussage All Saints Rural 1 – – – – 1 Dinwiddy, 2009 Little Keep Urban/rural 17 9 – – 3 29 Putnam, 1998 Littlewood Farm Urban/rural 1 – – – – 1 Farwell and Molleson, 1993 Camp Urban/rural 33 23 3 – 54 113 Davies and Grieve (1986) Poundbury Pipeline Urban/rural – 3 1 – 1 5 Davies and Thompson, 1987 Southfield House Urban – 1 – – – 1 Graham et al., 2007 Tarrant Hinton Rural – – – – 8 8 Hearne et al., 1999 Tolpuddle Bypass Rural 2 2 – – 1 5 Leonard, 2008 Wyke Regis Rural – 2 – 1 – 3 Total 99 63 10 2 117 291

of biases in historical urban development, most have determine age and sex because of preservation factors, been identified in the southeast of but have not burial truncation, and skeletal completeness, which been excavated in their entirety. In the countryside, bur- affect the demographic results obtained (Walker, 1995; ial areas have been identified in environs of villas, asso- Bello et al., 2006). The limitations identified by Pearce ciated with small settlements or military establishments (2008) above have influenced the samples available from (Cleary, 1987, 2000; Pearce, 2008). In Roman Britain, this region [see Redfern and DeWitte (2011)]. The major- cremation and inhumation burials have been identified, ity of individuals have been excavated from the environs but it seems that inhumation was more typical in this of Dorchester, in response to road building and expan- province of the Empire (Philpott, 1991). According to sion of the town in later periods. As Table 1 shows, the Philpott (1991), the majority of individuals were buried majority of cemeteries have a rural character, and there in a container in an extended position, often accompa- are intersite differences in the age and sex of the extant nied by grave-goods. He observes that although the population. Demographic analysis of the sample shows, range and type of grave-goods are subject to intense re- however, that it conforms to an attritional profile as gional and temporal variation, which in its simplest there are a greater number of young subadults, middle, terms reflects variation in population origin and beliefs, and older adult individuals (see Figure 1), and the male- different status groups are evident based on the type of to-female ratio of 1.5:1 is within normal limits (Cham- burial container (Philpott, 1991). According to Struck berlain, 2006). (2000), in Britain (Britannia), burial status appears to Formal burial grounds were quickly adopted around differ according to cemetery type. Struck’s (2000) the major settlement Durnovaria, in the Dorset region of research has shown that only 3% of all first to third cen- southwest England (Esmonde Cleary, 1987). Individuals tury AD graves show elite characteristics, the majority were buried in extended ‘‘supine’’ positions, although (2.5%) of which were discovered at city cemeteries, with prone and decapitation rites were also practiced (Leech, smaller towns accounting for just 0.5% of the total. Her 1980). The majority of individuals were buried in coffins analysis also demonstrates that children and females made of wood, lead, and stone, and a small number of were not excluded from high status burials because of individuals were also placed in other sarcophagi and/or their own rank and wealth or association with elite indi- interred in a frescoed mausoleum (e.g., Poundbury viduals (Struck, 2000). Camp) (Leech, 1980; Smith et al., 1997). Toller’s (1977) Many studies of Romano-British cemeteries report sex study of lead coffins in Britain shows that their distribu- and age differences in the samples available for study. tion conforms to the distribution of wealth in Britannia, Urban cemeteries appear to have a paucity of female with the majority uncovered at urban centers, with Dur- burials (Davison, 2000; Crowe, 2001), but many of these novaria having the second highest number of lead coffins cemeteries were strongly associated with the military excavated from its cemeteries. (e.g., , Essex), and, therefore, a higher number In the Romano-British period, people were buried with of male burials should not be unexpected. The low recov- a range of grave-goods (e.g., pottery), but the range of ery of infant burials (3 years old) from formal ceme- items expanded over time (Hamlin, 2007). Hamlin’s teries is believed to have been caused by age-related (2007) review of funerary evidence from the region trends in funerary practices, whereby these individuals shows that gender was not the determining factor in were preferentially buried in settlements (Scott, 1999). whether a person received grave-goods in their burial, The burial of infants also displays temporal variation, but intercemetery differences in the number of grave- with greater numbers being buried in cemeteries during goods given to either sex were observed, as was intrasex the third and fourth centuries AD (Pearce, 2001). The variation in the number of grave-goods. Interestingly, no cemetery populations available for study are also influ- gendered artifact classes (e.g., currency and personal enced by the geographic and temporal biases outlined ornament) were present in this period, but a number of earlier (Esmonde Cleary, 2000; Pearce, 2008), in addition items were specific to males/females (Hamlin, 2007). to limitations placed on the bioarchaeologist’s ability to Hamlin’s (2007) research has also shown that there is a

American Journal of Physical Anthropology 200 R.C. REDFERN AND S.N. DEWITTE

Fig. 1. Demography of the sample used in the study by age and sex group. strong age-related (subadult/adult) trend in the disposi- unknown, no coffin, wood coffin, lead-lined wood coffin, tion of grave-goods, which is proposed to relate to con- lead coffin, and stone coffin. cepts of the Roman life course. In the Romano-British pe- riod, coffin burials were the most frequent choice for Age and sex estimation males and females (92.6% of the 737 burials in Hamlin’s study). Age-related differences in coffin use were Subadults ( 20 years) were aged using a combination of observed, with adults more likely to receive a coffin burial dental eruption, diaphyseal length, and epiphyseal fusion than subadults; unlined wooden coffins were the most fre- methodologies (Ubelaker, 1989; Scheuer and Black, 2000). quently used type in Dorset, with a few individuals bur- When both dental and skeletal estimates were available, ied in lead-lined wood, stone, and stone with wood. Inter- dental age was used, because this has the strongest corre- estingly, there was no statistically significant difference in lation to chronological age and provides the most reliable the use of unlined wooden coffins for subadult or adult estimates (Lewis and Garn, 1960). Age-at-death in adults internments, indicating that coffin type was not related to was determined using degeneration of the pubic symphy- the age of the deceased (Hamlin, 2007; Table 2). sis and iliac auricular surfaces and sternal rib end mor- phology (Isc_ ¸an and Loth, 1986a,b; Buikstra and Ubelaker, 1994). Sex was estimated using morphology of the skull MATERIALS AND METHODS and pelvis (Buikstra and Ubelaker, 1994), and the sex and The individuals included in this present study date age-group divisions devised by Buikstra and Ubelaker from the first to early fifth centuries AD and are pre- (1994) were used. dominantly derived from the environs of Dorchester [see figures presented in Redfern and DeWitte (2011)]. The Effect of burial type on risk of mortality cemetery samples reflect a long history of excavation To determine the effect of status on risk of mortality work in the county, from the Victorian period to the across the lifespan, burial type was used as a proxy for present-day, and therefore it is not possible to break the status and modeled as a covariate affecting the parame- sample down into more discrete time periods (Redfern, ters of the Siler model of mortality. The Siler model is a 2006, 2007, 2008, 2010). Using the sample previously parsimonious five-parameter model of mortality that fits published by the authors (Redfern and DeWitte, 2011), a a wide range of human mortality patterns (Siler, 1979; total of 291 individuals were selected for the current Gage, 1988). study based on the availability of reliable archaeological information about burial type and that their skeletons b b ðÞ¼a 1a þ a þ a 3a were in articulation at the time of excavation: 116 suba- ha 1e 2 3e dults (\20 years old), 98 adult males, 63 adult females, 2b1a 10 ambiguous sex, and 2 indeterminate individuals. All The first component of the Siler model, a1e , repre- these individuals were between stages zero (excellent) to sents juvenile mortality, which typically is very high at three (weathered compact bone) of preservation (Buik- birth and then decreases rapidly with age; the a1 param- stra and Ubelaker, 1994; Tables 1 and 2). eter specifies the risk associated with infant and child- The use of coffins in Roman Dorset has been described hood causes of death, and b1 specifies the rate at which earlier, and, in this study, we used the following divi- this risk changes with age. The second component of the sions based on the work of Hamlin (2007) as a gauge of model, a2, is an age-independent component that is status: no coffin, wood coffin, lead-lined wood coffin, and associated with causes of death that are unrelated to an lead coffin; the highest status burials were within mau- individual’s age (e.g., accidental causes of death). The b3a soleums with people being interred in coffin type last component of the Siler model, a3e , represents the

American Journal of Physical Anthropology STATUS AND HEALTH IN ROMAN DORSET 201

TABLE 2. Variation in skeletal preservation (stages 0–3) displayed by coffin type, age group, and sex Coffin variable Age group Sex 0 1 2 3 No coffin Fetal – 31 – – – Infant – 13 7 1 – Child – 31–– Adolescent – 2––– Young adult Male 7––– Female 2––– Ambiguous sex 1––– Middle adult Male 12 1 1 – Female 7––– Old adult Male 5––– Female 5––– Adult Male 3––– Female 1––– Ambiguous sex – 1 – – Undetermined sex – 1 – – Wood Infant – 12 8 – – Child – 13 4 1 – Adolescent – 12 2 – – Young adult Male 17 5 4 2 Female 1131– Ambiguous sex 1––– Middle adult Male 18 3 – – Female 10 2 – – Ambiguous sex 1––– Old adult Male 2–1– Female 5––– Adult Male 3133 Female 351– Ambiguous sex 113– Undetermined sex – 1 – – Lead Middle adult Male 1––– Lead-lined wood Child – 1––– Mausoleum: no coffin Infant – 2––– Mausoleum: wood Child – 1––– Young adult Male 2––– Middle adult Male 2––– Female 1––– Old adult Female 1––– Mausoleum: lead Young adult Male – 1 – – Female 1––– Middle adult Male – 1 – – Mausoleum: coffin unknown Infant – 2––– Mausoleum: lead-lined wood Infant – – 1 – – Mausoleum: stone Young adult Male 1––– Middle adult Female – – 1 –

typical senescent pattern of mortality, which is low at Given that slaves were often buried in the mausolea of birth and younger ages, but increases with adult age; a3 their owners (Toynbee, 1996), it is possible that the high- specifies the risk associated with senescent causes of status burial sample includes some low status people; death, and b3 specifies the rate at which this risk the potential effect of this is discussed below. changes with age. The three components of the Siler Given our previous findings that the risk of mortality model are independent of one another, and so surviving was elevated for males compared to females in Britannia one component has no influence on risk of mortality dur- and results from a recent study indicating sex differen- ing another component (Wood et al., 2002). tials in the effects of physiological stress on mortality Burial type was modeled as a covariate affecting the risk in an archaeological sample from Britain (DeWitte, Siler model in two ways. The first approach was to 2010), we also examined sex differences in the effect of model burial type as a covariate affecting the entire Siler status on risks of death for this study. We first confirmed model (i.e., burial type was modeled as proportional to that sex did, in fact, affect risk of mortality in this par- the entire hazard, independent of age). However, the ticular sample (as the sample composition differs slightly effects of status on risks of mortality might not be uni- from that used in our previous study) by modeling sex as formly distributed across all ages, and the aggregate pat- a covariate affecting the Gompertz–Makeham model of tern may mask important subpopulation differences in mortality (males were scored as 1 and females were risks of mortality (Vaupel and Yashin, 1985; Wood et al., scored as 0 for this analysis). Then, to assess sex differ- 1992). Therefore, we also modeled burial type as a cova- ences in the effect of status on risk of death, we modeled riate affecting the juvenile and senescent components of burial type as a covariate affecting the Gompertz–Make- the Siler model independently to allow for some varia- ham model and compared the results for males versus tion with age in the effect of status on risk of mortality. females. The Gompertz–Makeham model includes the

American Journal of Physical Anthropology 202 R.C. REDFERN AND S.N. DEWITTE

TABLE 3. The effect of the burial type covariate on the entire TABLE 4. The effect of the burial type covariate on the Gom- Siler hazard of mortality and on the juvenile and senescent com- pertz–Makeham hazard of mortality for men and women in the ponents of the Siler model for the total Romano-British sample total Romano-British sample Siler hazard Juvenile mortality Senescent mortality Men Women 20.098 (20.36, 0.14) 21.47 (22.45, 20.87) 0.13 (20.20, 0.36) 20.32 (20.72, 0.11) 20.19 (20.72, 0.35)

The 95% confidence intervals are shown in parentheses. The 95% confidence intervals are shown in parentheses. second and third components (i.e., the age-independent and senescent components) of the Siler model shown RESULTS above (Wood et al., 2002): The estimated values and 95% confidence intervals of the parameters representing the effect of the burial-type ba hðaÞ¼a1 þ a2e covariate on the entire Siler model, independent of age, and on the juvenile and senescent components of the For all of the analyses described earlier, individuals who model for the total Romano-British sample are shown in were not buried in coffins were assigned a value of 0, Table 3. For the model with burial type specified as a and individuals buried in any type of coffin or in mauso- covariate affecting the entire Siler hazard independent leums were assigned a value of 1. Parameters were of age, the estimated value of the parameter represent- estimated using maximum likelihood analysis with the ing the effect of the burial type covariate is not signifi- program mle (Holman, 2005). A significant negative cantly different from zero. This suggests that status had estimate for the parameter representing the effect of the no significant effect on the risk of mortality when all burial type covariate would indicate that the risk for ages are assessed simultaneously. However, when mod- people buried in coffins was lower than that for people eled as a covariate affecting the juvenile and senescent without coffins. components of the hazard independently, the estimated As shown in Table 1, the cemetery sample from Dorset value of the burial type covariate effect on juvenile mor- used in this study includes urban and rural cemeteries tality is significantly less than zero, but the estimated and those that have both urban and rural characteristics effect on senescent mortality is not significantly different (urban/rural). Because of this variation in cemetery con- from zero. These results suggest that status had a signif- texts, we assessed the effect of status on the risk of mor- icant effect on childhood mortality (i.e., higher status tality not only for the combined sample (i.e., all ceme- was associated with lower risks of mortality for infants teries shown in Table 1), but also separately for the and young children), but not senescent mortality. urban, rural, and urban/rural subsamples to determine Analysis of the effect of the sex covariate on risk of if there were any differences among the three. Compari- mortality indicated that males were at higher risks of son of the effects of status on mortality for males and death than females, similar to our previous findings (the females was limited to the combined sample with all con- estimated effect of the sex covariate was 0.44, with a texts included, as there were insufficient sample sizes 95% confidence interval of 0.09–0.77). Table 4 shows the from any of the subsamples to assess the difference estimates of the effect of burial type on risk of mortality between the sexes within each subsample. for males and females in the total sample. For both One major advantage of the Gompertz–Makeham and sexes, the estimated values of the effect of the burial Siler models is that they require the estimation of a type covariate are not significantly different from zero; small number of parameters and thus can be applied to these results are consistent with the result for all adults small samples, as they smooth the random variation in combined and suggest that status did not significantly mortality data that is an artifact of small samples with- affect the risk of mortality for males or females in this out imposing any particular age pattern on the data sample. (Gage, 1988). However, this efficient use of small sample The results for the urban, rural, and urban/rural cem- sizes has the drawback of potentially masking underly- eteries are shown in Table 5. The results from the rural ing heterogeneity in risks of death. By including covari- and urban/rural sites are consistent with those from the ates in our models, we can examine the effects of some total sample, that is, burial type has no significant effect potential causes of heterogeneity (i.e., social status and on risk of mortality when it is modeled independent of sex). It is possible that there are other sources of hetero- age nor does it have a significant effect on the senescent geneity in risks of mortality (e.g., nutritional status and component; however, coffin/mausoleum burial is associ- geographic origin) that we are not capturing. Analysis of ated with significantly reduced risks of mortality in such factors is beyond the scope of the current study, as childhood. The results from the urban subsample differ the addition of more covariates necessitates a larger from those observed in the other subsamples and from sample size than is currently available. the total sample. For the urban subsample, the esti-

TABLE 5. The effect of the burial type covariate on the entire Siler hazard of mortality and on the juvenile and senescent compo- nents of the Siler model in the urban, rural, and urban/rural subsamples Location Siler hazard Juvenile mortality Senescent mortality Urban 21.94 (22.82, 21.25) 22.51 (23.39, 21.83) 21.42 (22.33, 20.73) Rural 0.08 (20.40, 0.50) 21.59 (210, 20.33) 20.07 (20.55, 0.34) Urban/Rural 0.32 (20.04, 0.64) 20.87 (21.65, 20.12) 0.24 (20.11, 0.56)

The 95% confidence intervals are shown in parentheses.

American Journal of Physical Anthropology STATUS AND HEALTH IN ROMAN DORSET 203 mated value of the parameter representing the burial mortality, and the true difference in risk of death type covariate is significantly lower than zero when bur- between high and low status individuals might be even ial type is modeled on the Siler hazard independent of greater (in favor of high status people) than our esti- age and when it is modeled as a covariate affecting juve- mates reveal. nile and senescent mortality separately. These results The significantly reduced risk of death associated with suggest that in the urban populations, higher status was coffin/mausoleum burial for juveniles, but not for adults associated with significantly reduced risks of death in the rural and urban/rural samples might reflect the across all ages and not just among children as in the effects of strong selective mortality during childhood in other subsamples. those populations. Mortality tends to be selective and target individuals with higher frailty, which is an indi- vidual’s relative risk of dying compared to other people DISCUSSION in the same population (Vaupel et al., 1979). In all con- The findings of this study challenge many current texts, in Romano-British Dorset, high status likely interpretations about status and funerary treatment in exerted strong protective effects on children, such that Britannia. It is considered that the discrepancy between higher status children had lower frailty compared to these and other findings arises from bioarchaeological their lower status peers. If mortality was strongly selec- data, and, in particular, mortality modeling, being tive with respect to frailty in childhood in these popula- excluded from funerary and status studies, because they tions, the individuals who survived to older ages would are deemed unreliable or of no value (Gowland, 2006). have had lower average frailty than the original cohort Our findings show that there may be disparities between exposed to childhood mortality. By selecting out of the what is ‘‘read’’ from the funerary context and the population those individuals with highest frailty, a large remains of the person contained within, particularly proportion of which would likely have been among the with respect to a lack of intrastatus differences in each lower statuses, selective mortality could have resulted in sex. We accept that this result may reflect our choice of reduced apparent differences in the risks of mortality coffin type as an indicator of status and the widespread between high and lower status people during adulthood. use of unlined wooden coffins in Roman Dorset. Never- That is, during childhood, there might have been signifi- theless, the result indicating that burial type had no sig- cant differences in frailty between people of high and nificant effect on risk of death for both sexes does con- low status, but strong selection against low status chil- form to the funerary evidence that shows considerable dren with high frailty could have resulted in reduced dif- overlap in practices between males and females and the ferences in frailty between high and low status adults in lack of gendered artifact grave-good classes. Consequen- these populations. tially, our results support Hamlin’s (2007) findings, but Status as a buffer for child mortality provides a more also those in funerary research more generally, in that nuanced understanding of our previous findings, which the grave context is a not a true reflection of the indicated that subadults had a higher mortality risk in deceased, because it is a sociocultural construction (e.g., Dorset during this period (Redfern and DeWitte, 2011). Pearce, 2010). This finding is also important, because, in Storey’s (1992) analysis of preindustrial health in Mexico part, it lends weight to Struck’s (2000) hypothesis that found that compared to rural groups, urban populations the number of elites in Roman Britain was very small. had higher mortality rates in infant and child age- These new findings also support the results of our pre- groups, particularly for lower socioeconomic groups. vious study that Romanization increased male mortality Therefore, the role of status in buffering is not unex- risk compared to females. As discussed in Redfern and pected (Panter-Brick, 1998), particularly as the funerary DeWitte (2011), this result is contrary to the increased evidence shows that although coffin type was not related cultural buffering afforded by having a male body in the to age, adults unlike subadults were more likely to be Roman world (Foxhall and Salmon, 1998). We believe buried in a coffin (Hamlin, 2007). Our new findings sup- that this cultural advantage may not have been suffi- port studies in other archaeological populations that cient protection from environmental stressors associated have demonstrated that high-status samples have lower with employment, particularly for nonelite males, and subadult mortality (Cowie et al., 2008); however, this is conclude that underlying sex-differences account for this the first to do so in the Roman Empire. The survival of finding. subadults is determined by their immediate environment The estimated values of the parameter representing and care-givers (Saunders and Barrans, 1999; Lewis, the effect of the burial type covariate shown above sug- 2007), and in higher-status environments, several buf- gest that in this region of Britannia, in general, and in fers may have afforded them better protection. In the rural and urban/rural contexts, in particular, status higher-status homes, environmental hazards may have had significant effects on the risks of mortality for been reduced because of plumbing and the use of infants and children, but it did not significantly affect latrines, which would have made living conditions more risks of mortality for adults. Higher status seems to sanitary (Scobie, 1986; Jackson, 2000), although we can have been associated with lower risks of mortality for never be certain about standards of hygiene. Many of children throughout Roman Dorset, but in the rural and these homes also had water delivered from the aqueduct, urban/rural areas, status did not apparently strongly which would have reduced the risk of contamination affect the risk of death for adults. The pattern from the (Howe, 1997) compared to lower-status homes, which urban cemeteries differs from that observed in the rural relied on well water, as garbage pits were often dug near and urban/rural subsamples, as higher status appears to wells (Jackson, 2000). have been associated with significantly lower risks of Stable isotope data show that migrants, particularly mortality for adults in addition to children in urban from the Mediterranean (including one child who lived areas. As mentioned earlier, it is possible that there are in Durnovaria), and different food-ways and dietary some low status people buried in mausolea; if so, we practices were present in the region (Richards et al., might have underestimated the effect of status on risk of 1998; Redfern et al., 2010). Richards et al.’s (1998) study

American Journal of Physical Anthropology 204 R.C. REDFERN AND S.N. DEWITTE of Poundbury Camp found that individuals buried in but fault was assigned to the mother for behaving badly high status burials (lead coffins and mausoleums) con- and or neglecting her child (Bradley, 2005). The results sumed more marine products compared to those in of this study show that although high-status living was wooden coffins. However, this finding has not been not without its problems, for the majority of subadults, observed in other cemeteries in Dorset (Redfern, unpub- it was of some benefit. lished data) and, therefore, the role of diet as a status The differences observed between the urban subsam- buffer is not as clear. Archaeobotanical records from ple and the rural and urban/rural subsamples (i.e., sta- Roman Britain show that there is no clear correlation tus affected risk of mortality for all ages in urban areas, between location of a settlement (i.e., rural vs. urban in but only for children in other locations) might indicate character) and the types of food consumed and that that people in towns were, in general, better buffered access to new foods in rural sites and small towns was against physiological stress, at least during childhood, not limited to high-status groups (van der Veen et al., than people living in rural areas. In other words, the 2008). observed differences in the effect of status on mortality Stable isotope analysis of subadult diet and weaning risk for adults suggest that mortality was not as strongly in Dorset shows that, as with other areas of the Empire, selective during childhood in urban areas as it was in children were weaned by the ages of 2–3 years (e.g., rural areas. A reduction in the strength of selective mor- Dupras and Tocheri, 2007; Prowse et al., 2008; Redfern tality might have allowed more urban, low-status indi- et al., in review). Other evidence for subadult diets is viduals with high frailty to survive to adulthood than present, including feeding bottles (Fildes, 1985), and, they would have had they lived in rural areas. This, in from excavations in Durnovaria, we know that a wet turn, might have resulted in the maintenance of greater nurse lived in the civitas, as a flagon, has the inscription variation in frailty throughout the lifespan in urban ‘‘NUTRICIS’’–property of Nutrix (Tomlin, 1993: 284). areas compared to rural areas and thus the maintenance Roman medical texts, such as Soranus and Celsus, rec- of significant differences in risks of mortality between ommended that weaning took place between the ages of high and low status people from childhood through 3 months and 3 years (Rawson, 2003: 7–30, 126). Wean- adulthood in urban areas. To a certain extent, this chal- ing is a risky process that had a high morbidity risk in lenges conventional wisdom, as preindustrial urban pop- past populations, because the immunity and protection ulations are considered to have had poor health (Storey, of breast milk are removed as new foods are introduced 1992; Paine and Storey, 2006; Scheidel, 2009, 2010), and that may have been made with contaminated water or there is a overlap in the range and type of pathogens were of insufficient nutritional quality (Katzenberg and types of disease (e.g., tuberculosis, rickets, and et al., 1996; Herring et al., 1998). Roman medical texts scurvy) occurring in both rural and urban settlements recommended that children were fed cereals—which is (Lewis, 2003). attested isotopically (Prowse et al., 2005, 2008)—but this In Roman Britain, an earlier study by Redfern and may have increased their risk of developing metabolic Roberts (2005) found that urban populations had higher diseases, because such foods may not have been supple- rates of infectious and metabolic diseases. Therefore, we mented with sources of vitamins C and D. In Roman must suppose that in Dorset, the buffering was cultural Dorset, rickets and scurvy have been observed in these rather than environmental. The benefits of living in Dur- age groups, and examination of their funerary contexts novaria compared to rural areas may have been from the shows that the majority was buried in coffins and one in amenities offered in the town, such as the aqueduct and a mausoleum, indicating that they were of high status access to a bathhouse (Putnam, 2007). Additionally, life in (Redfern, 2007). Consequentially, we can conclude that rural areas carries the risks of diseases and injury from although status did afford some protection, it was not agricultural work, such as becoming infected by patho- absolute. gens in the soil or from animals, the often debilitating As Roman Britain contained a diverse population, it is effects of heavy manual labor, and accidents sustained in very unlikely that one style of child-rearing was prac- animal husbandry or from agricultural tools (El Batawi, ticed, and, therefore, there may have been differences in 2003; Lewis, 2003). Unfortunately, very little is known this form of cultural buffering (Schell and Ulijaszek, about rural life in Roman Britain, particularly about the 1999). Woolf’s (2005) research into family life in the people themselves. This arises from very few rural ceme- northwest provinces of the Empire has shown that little teries being excavated in Britain (Pearce, 2008) and the epigraphic evidence is available, but suggests that the majority of research focusing on economic and settlement nature of the family and its relationship to wider kinship activities (Taylor, 2001). The possibly higher mortality and power networks would have been subject to consid- risk of rural populations is considered to reflect two out- erable regional differences, in addition to variation in comes; first, it is generally proposed that the majority of the adoption/adaptation of Roman culture and concepts rural dwellers in this period were more likely to be indig- of family life (Rawson, 1986a, 2003; Revell, 2005). The enous (Birley, 1964, 1979; Mattingly, 2006), and, there- Roman medical texts recommended a range of child-rear- fore, their interactions with newcomers may have ing practices, which were aimed at high-status groups in increased their risk of being exposed to new pathogens; the Mediterranean (Rawson, 2003; Bradley, 2005). As additionally, recent environmental archaeological sug- many inhabitants of Durnovaria were from this area, we gests that livestock was being imported (Albarella et al., must allow for the possibility that such practices were 2008), which may have brought with them diseases hith- followed in Britannia [see Gowland and Redfern (2010)]. erto not encountered. Stable isotope analysis of rural pop- Unfortunately, the information in these texts did not ulations from Dorset shows that they are more diverse make for the best health outcomes, as they recom- than previously expected, as different food-ways are pres- mended the withholding of colostrum (Celsus, 1935; ent (Redfern et al., 2010). This may reflect villa owners Holman, 1998). These practices would have increased importing slaves to work the land, whose health status morbidity and the risk of developing metabolic dis- may have been compromised by migration and their eases—indeed, this was recognized by Roman doctors, social status.

American Journal of Physical Anthropology STATUS AND HEALTH IN ROMAN DORSET 205 We must also consider the possibility that the differen- and encouragement and also thanks Simon Mays, AC ces observed between the urban subsample and the rural Archaeology, AOC Archaeology, Jacqueline McKinley and and urban/rural subsamples reflect stronger mortality Wessex Archaeology, Peter Woodward and Dorset County among adults in urban cemeteries than in rural areas, Museum, Dorset Natural History and Archaeological Soci- because there were causes of death in urban areas that ety, The Natural History Museum (London), the Duck- did not exist in rural areas, which would have targeted worth Laboratory (Cambridge University), and The Priest’s adults and selected out low-status individuals (Harrison House Museum (Wimborne Minster). S.N.D. thanks Eric and Gibson, 1978). Low-status people may have had Jones for his helpful comments on the manuscript. increased risk, because their living conditions were more unsanitary, as they could not afford measures such as drainage. Their employment may have also led to health problems, such as exposure to pathogens during butch- LITERATURE CITED ery, pollution from metalworking, and being used in Adkins L, Adkins RA. 1998. Handbook to life in Ancient Rome. occupations that would have increased their risk of dis- New : Oxford University Press. ease (Lewis, 2003), such as working as a cleaner in a Albarella U, Johnstone C, Vickers K. 2008. The development of bathhouse. Furthermore, the likely greater number of animal husbandry from the late Iron Age to the end of the migrants in urban areas (Birley, 1979; Evans et al., Roman period: a case study from south-east England. J 2006) might have resulted in a greater proportion of frail Archaeol Sci 35:1828–1848. adults in urban versus rural areas. Migrants from the Bello SM, Thomann A, Signoli M, Dutour O, Andrews P. 2006. Age and sex biases in the reconstruction of past population Mediterranean into urban areas, for example, may have structures. Am J Phys Anthropol 129:24–38. been at increased mortality risk, because they were Birley A. 1964. Life in Roman Britain. London: BT Batsford Ltd. exposed to pathogens in Roman Dorset to which they Birley A. 1979. The people of Roman Britain. London: BT Bats- lacked immunity and their biological and immunological ford Ltd. adaptations may have been made to environmental con- Bradley KR. 2005. The Roman child in sickness and health. In: ditions very different from those in Britain. However, George M, editor. The Roman family in the Empire. Oxford: the cultural buffering associated with high status would Oxford University Press. p 67–92. presumably have benefitted high-status migrants as it Buikstra JE, Ubelaker DH. 1994. Standards for data collection did high-status native inhabitants. from human skeletal remains. Proceedings of a seminar at the Field Museum of Natural History organized by Jonathan Haas. Arkanas: Arkansas Archaeological Survey Research Se- CONCLUSION ries No 44. Buzon M, Judd MA. 2008. Investigating health at Kerma: sacri- Our study of the relationship between status and mor- ficial versus nonsacrificial individuals. Am J Phys Anthropol tality risk of a region in Roman Britain reveals that 136:93–99. high-status individuals, particularly children, had a Carroll M. 2006. Spirits of the dead: Roman funerary commemo- lower mortality risk compared to lower-status people; ration in Western Europe. Oxford: Oxford University Press. and for those buried in urban cemeteries, higher-status Celsus. 1935. De medicina. London: William Heinemann. Chamberlain A. 2006. Demography in archaeology. Cambridge: individuals of all age-groups had a lower mortality risk. Cambridge University Press. These results provide further insights into our earlier Cowie R, Bekvalac J, Kausmally T. 2008. Late 17th to 19th cen- findings, which reveal that Romanization had a deleteri- tury burial and earlier occupation at All Saints, Chelsea Old ous effect on health because of the introduction of urban- Church, Royal Borough of Kensington and Chelsea. London: ism and increased population migration. Museum of London, MoLAS Archaeology Studies 18. We propose that the cultural buffering afforded by Crowe F. 2001. Women, burial data and issues of inclusion. The being of high-status enabled people to more effectively problems and potentials of Romano-British cemeteries. In: deal with urban environments and migration, with Dixon S, editor. Childhood, class and kin in the Roman world. lower-status individuals having greater risk because of London: Routledge. p 144–162. Davies SM, Bellamy PS, Heaton MJ, Woodward PJ. 2002. Exca- their forms of employment and living conditions. Our vations at Alington Avenue, Fordington, Dorchester, Dorset, findings about the relationship between urban living and 1984–87. Dorchester: Dorset Natural History and Archaeologi- mortality risk are not clear, because urbanism is known cal Society Monograph Series Number 15. to be a risk to health (Schell, 1997; Fotso, 2006; O’Reilly Davies SM, Grieve D. 1986. The Poundbury pipeline: archaeo- et al., 2007), but it appears to have conferred an advant- logical observations and excavations. Proc Dorset Nat Hist age in Roman Dorset. The reasons for this are not clear, Archeol Soc 107:81–88. as access to healthcare and sanitation was likely to be Davies SM, Thompson CN. 1987. Archaeological evaluation at status-related. It is hoped that future studies in other Southfield House, Dorchester, Dorset. Proc Dorset Nat Hist Roman locales will be able to help clarify this finding. Archeol Soc 107:101–110. Overall, our research has demonstrated that a bioarch- Davison C. 2000. Gender imbalances in Romano-British ceme- tery populations: a re-evaluation of the evidence. In: Pearce J, aeological examination of funerary status shows that the Millett M, Struck M, editors. Burial, society and context in relationship between the burial evidence and the body is the Roman world. Oxford: Oxbow Books. p 231–237. more nuanced and complex than previously thought, and DeWitte SN. 2010. Sex differentials in frailty in medieval Eng- it provides new insights into life in Roman Britain. land. Am J Phys Anthropol 143:285–297. Dinwiddy KE. 2009. A late Roman cemetery at Little Keep, Dorchester, Dorset. Salisbury: Wessex Archaeology.http:// ACKNOWLEDGMENTS www.wessexarch.co.uk/files/Little_Keep_Dorchester_64913.pdf [Accessed 01 July 2010]. Both authors thank Chris Hamlin for allowing us to use Dupras TL, Tocheri MW. 2007. Reconstructing infant weaning the results of her doctoral research in our study and for histories at Roman period Kellis. Egypt using stable isotope helping with the funerary information. R.C.R. thanks analysis of dentition. Am J Phys Anthropol 134:63–74. Jenny Hall, Roy Stephenson, and her doctoral tutors El Batawi MA. 2003. Health of workers in agriculture. Cairo: Megan Brickley and Simon Esmonde Cleary for their help WHO Regional Publications, Eastern Mediterranean Series 25.

American Journal of Physical Anthropology 206 R.C. REDFERN AND S.N. DEWITTE

Esmonde Cleary SE. 1985. The quick and the dead: suburbs, Holman DJ. 2005. mle: a programming language for building cemeteries, and the town. In: Grew F, Hobley B, editors. likelihood models. Version 2.1 ed. Seattle, WA. Roman urban topography in Britain and the western Empire. Holman SR. 1998. Infant feeding in Roman antiquity: prescrip- York: British Archaeology Research Report 59. p 74–77. tion and risk. Nutr Today 33:113–120. Esmonde Cleary S. 1987. Extra-mural areas of Romano-British Hope V. 2009. Roman death. The dying and the dead in ancient towns. Oxford: British Archaeological Reports, British Series Rome. London: Continuum. 169. Howe GM. 1997. People, environment, disease and death. A Esmonde Cleary S. 2000. Putting the dead in their place: burial medical geography of Britain throughout the ages. Cardiff: location in Roman Britain. In: Pearce J, Millett M, and Struck University of Wales Press. M, editors. Burial, society and context in the Roman world. Isc_ ¸an M, Loth S. 1986a. Determination of age from the sternal Oxford: Oxbow Books. p 127–142. rib in white males: a test of the phase method. J Forensic Sci Evans J, Stoodley N, Chenery C. 2006. A strontium and oxygen 31:122–132. isotope assessment of a possible fourth century immigrant Isc_ ¸an M, Loth S. 1986b. Determination of age from the sternal population in a cemetery, southern England. J rib in white females: a test of the phase method. J Forens Sci Archaeol Sci 33:365–272. 31:990–999. Farwell DE, Molleson TL. 1993. Excavations at Poundbury Jackson R. 2000. Doctors and diseases in the Roman Empire. 1966–80, Vol.II: the cemeteries. Dorchester: Dorset Natural London: British Museum Press. History and Archaeological Society Monograph Series Number Joshel SR. 2010. Slavery in the Roman world. Cambridge: Cam- 11. bridge University Press. Fildes V. 1985. Breasts, bottles and babies: a history of infant Katzenberg MA, Herring DA, Saunders SR. 1996. Weaning and feeding. Edinburgh: Edinburgh University Press. infant mortality: evaluating the skeletal evidence. Yrbk Phys Flemming R. 2000. Medicine and the making of Roman women: Anthropol 39:177–199. gender, nature and authority from Celsus and Galen. Oxford: Langley RL. 2003. Sex and gender differences in health and dis- Oxford University Press. ease. Durham, NC: Carolina Academic Press. Fotso JC. 2006. Child health inequalities in developing coun- Lao O, Lu TT, Nothnagel M, Junge O, Freitag-Wolf S, Caliebe tries: differences across urban and rural areas. Int J Equity A, Balascakova M, Bertranpetit J, Bindoff LA, Comas D, Health 5:9. Holmlund G, Kouvatsi A, Macek M, Mollet I, Parson W, Palo Foxhall L, Salmon J, editors. 1998. When men were men. Mas- J, Ploski R, Sajantila A, Tagliabracci A, Gether U, Werge T, culinity, power and identity in Classical antiquity. London: Rivadeneira F, Hofman A, Uitterlinden AG, Gieger C, Wich- Routledge. mann HE, Ru¨ ther A, Schreiber S, Becker C, Nu¨ rnberg P, Nel- Gage TB. 1988. Mathematical hazard models of mortality: an al- son MR, Krawczak M, Kayser M. 2008. Correlation between ternative to model life tables. Am J Phys Anthropol 76:429– genetic and geographic structure in Europe. Curr Biol 441. 18:1241–1248. Garnsey P. 1968. Legal privilege in the Roman Empire. Past Leach S, Eckhardt H, Chenery C, Muldner G, Lewis M. 2010. A Present 41:3–24. ‘lady’ of York: migration, ethnicity and identity in Roman Gosden C. 2006. Archaeology and colonialism. Cultural contact York. Antiquity 84:131–145. from 5000 BC to the present. Cambridge: Cambridge Univer- Leech R. 1980. Religion and burials in south and sity Press. north Dorset. In: Rodwell W, editor. Temples, churches and re- Gowland R. 2006. Ageing the past: examining age identity from ligion: recent research in Roman Britain. Oxford: BAR British funerary evidence. In: Gowland R, Knu¨ sel CJ, editors. Social Archaeological Reports. p 329–366. archaeology of funerary remains. Oxford: Oxbow Books. Leonard A. 2008. Late Iron Age and Romano-British burials p143–154. and associated activity at the former allotments, Church Gowland RL, Redfern RC. 2010. Childhood health in the Roman Knapp, Wyke Regis, Weymouth. Proc Dorset Nat Hist Archeol world: perspectives from the centre and margin of the Soc 129:115–126. Empire. Childhood Past: Int J 3:15–42. Lewis AB, Garn SM. 1960. The relationship between tooth forma- Graham A, editor. 2007. The excavation of five beaker burials, tion and other maturational factors. Angle Orthod 30:70–77. the Iron Age and Romano-British settlements and the 4th Lewis M. 2003. A comparison of health in past rural, urban and century courtyard villa at Barton Field, Tarrant Hinston, Dor- industrial England. In: Murphy P, PEJ, editors. The set, 1968–1984.Excavation by RG Tanner and AG Giles, Wim- environmental archaoelogy of industry. Symposia of the Asso- borne Archaeological Group. Dorchester: Dorset Natural His- ciation for Environmental Archaeology No 20. Oxford: Oxbow tory and Archaeological Society Monograph Series 17. Books. p 154–161. Green CS, Paterson M, Biek L. 1981. A Roman coffin-burial Lewis ME. 2007. The bioarchaeology of children. Cambridge: from the Crown building site. Dorchester: with particular ref- Cambridge University Press. erence to the head of well-preserved hair. Proc Dorset Nat Mattingly D. 2006. An imperial possession: Britain in the Hist Archeol Soc 103:67–94. Roman Empire. London: Penguin Books Ltd. Hamlin C. 2007. Material expression of social change: the mor- McElroy A, Townsend PK. 1996. Medical anthropology in ecolog- tuary correlates of gender and age in late Pre-Roman Iron ical perspective. Oxford: Westview Press. Age and Roman Dorset. [PhD dessertation]. Milwaukee: Uni- Nettle D. 2010. Why are there social gradients in preventative versity of Wisconsin-Milwaukee. health behavior? A perspective from behavioral ecology. PLoS Harlow M, Laurence R. 2002. Growing up and growing old in One 5:e13371. ancient Rome: a life course approach. London: Routledge. O’Reilly G, O’Reilly D, Rosato M, Connelly J. 2007. Urban and Harrison GA, Gibson JB, editors. 1978. Man in urban environ- rural variations in morbidity and mortality in Northern Ire- ments. Oxford: Oxford University Press. land. BMC Public Health 26:123. Hearne CM, Birkbeck V. editors, 1999. A35 Tolpuddle to Pud- Paine RR, Storey GR. 2006. Epidemics, age at death, and mor- dletown bypass DBFO, Dorset, 1996–8, incorporating excava- tality in Ancient Rome. In: Storey GR, editor. Urbanism in tions at Tolpuddle Ball. Salisbury: Trust for Wessex Archaeol- the preindustrial world: cross-cultural approaches. Tusca- ogy Report Number 15. loosa: The University of Alabama Press. p 69–85. Herring DA, Saunders SR, Katzenberg MA. 1998. Investigating Panter-Brick C, editor. 1998. Biosocial perspectives on children. the weaning process in past populations. Am J Phys Anthro- Cambridge: Cambridge University Press. pol 105:435–439. Pearce J. 2001. Infants, cemeteries and communities in the Hingley R. 1989. Rural settlement in Roman Britain. London: Roman provinces. In: Davies G, Gardner A, Lockyear K, edi- B.A. Seaby Ltd. tors. TRAC 2000 Proceedings of the Tenth Annual Theoretical Hingley R. 2005. Globalizing Roman culture. Unity, diversity Roman Archaeology Conference, London 2000. Oxford: Oxbow and Empire. London: Routledge. Hippocrates. Books. p125–142.

American Journal of Physical Anthropology STATUS AND HEALTH IN ROMAN DORSET 207

Pearce J. 2008. Burial evidence from Roman Britain. The un- Scheidel W. 2010. Physical wellbeing in the Roman world. numbered dead. Pour une archaeologie du rite. Nouvelles per- Princeton/Stanford Working Papers in Classics. spectives de l’archaeologie funeraire. Etudes reunies par John Schell LM. 1997. Culture as a stressor: a revised model of bio- Scheid. Paris: Collection de L’Ecole Francaise de Rome. p 29– cultural interaction. Am J Phys Anthropol 102:67–77. 42. Schell LM, Ulijaszek SJ. 1999. Urbanism, health and human Pearce J. 2010. Burial, identity and migrations in the Roman biology in industrialised countries. Cambridge: Cambridge world. In: Eckhardt H, editor. Roman diasporas: archaeologi- University Press. cal approaches to mobility and diversity in the Roman Scheuer L, Black S. 2000. Developmental juvenile osteology. Empire. Portsmouth, RI: Journal of Roman Archaeology Sup- London: Academic Press. plementary Series. p79–98. Scobie A. 1986. Slums, sanitation, and mortality in the Roman Philpott R. 1991. Burial practices in Roman Britain: a survey of world. Klio 68:399–433. grave treatment and furnishing AD 43–410. Oxford: British Scott E. 1999. The archaeology of infancy and infant death. Archaeological Reports British Series. Oxbow: British Archaeological Reports International Series Prowse TL, Saunders SR, Schwarcz HP, Garnsey P, Bondioli L, 819. Macchiarelli R. 2008. Isotopic and dental evidence for wean- Siler W. 1979. A competing-risk model for animal mortality. ing and young child feeding practices in an Imperial Roman Ecology 60:750–757. skeletal sample. Am J Phys Anthropol 137:294–308. Smith RJC, editor. 1993. Excavations at County Hall, Colliton Prowse TL, Schwarcz H, Saunders S, Bondioli L, Macchiarelli Park, Dorchester, Dorset, 1988. Salisbury: Wessex Archaeol- R. 2005. Isotopic evidence for age-related variation in diet ogy Report 4. from Isola Sacra, Italy. Am J Phys Anthropol 128:2–13. Smith RJC, Healy F, Allen MJ, Morris EL, Barnes M, Wood- Putnam B. 1998. Skeleton from Frampton. Unpublished report. ward PJ. 1997. Excavations along the route of the Dorchester Putnam B. 2007. Roman Dorset. Stroud: Sutton. by-pass, Dorset, 1986–8. Wessex Archaeology Report No. Dor- Rawson B. 1986a. Children in the Roman family. In: Rawson B, . editor. The family in ancient Rome: new perspectives. London: Startin DWA. 1982. Excavations at the Old Vicarage, Fording- Croom Helm. p170–200. ton, Dorchester, Dorset. Proc Dorset Nat Hist Archeol Soc Rawson B. 1986b. The Roman family. In: Rawson B, editor. The 103:43–61. Roman family: new perspectives. London: Croom Helm. p1–57. Steckel RH, Rose JC, editors. 2002. The backbone of history: Rawson B. 1991. Adult-child relationships in Roman society. In: health and nutrition in the Western Hemisphere. Cambridge: Rawson B, editor. Marriage, divorce, and children in ancient Cambridge University Press. Rome. Oxford: Clarendon Press. p 7–30. Stinson S. 1985. Sex differences in environmental sensitivity dur- Rawson B. 2003. Children and childhood in Roman Italy. ing growth and development. Yrbk Phys Anthropol 28:123–147. Oxford: Oxford University Press. Storey R. 1992. Preindustrial urban lifestyle and health. In: Redfern RC. 2006. A gendered analysis of health from the Iron Huss-Ashmore R, Schall J, Hediger M, editors. Health and Age to the end of the Romano-British period in Dorset, Eng- lifestyle change . MASCA Res Pap Sci Archaeol9:33–42. land (middle and late 8th century B.C. to the end of the 4th Struck M. 2000. High status burials in Roman Britain (first- century A.D.). Ph.D. dissertation, University of Birmingham. third century AD)—potential interpretation. In: Pearce J, Mil- Redfern RC. 2007. The influence of culture upon childhood: an lett M, Struck M, editors. Burial, society and context in the osteological study of Iron Age and Romano-British Dorset. In: Roman world. Oxford: Oxbow Books. p85–96. Harlow M, Laurence R, editors. Age and ageing in the Roman Taylor J. 2001. Rural society in Roman Britain. In: James S, Empire: Journal of Roman Archaeology Supplementary Se- Millett M, editors. Britons and Romans: advancing an ries. p 171–194. archaeological agenda. York: CBA Research Report 125, Coun- Redfern RC. 2008. A bioarchaeological investigation of cultural cil for British Archaeology. p46–59. change in Dorset, England (mid to late 4th century B.C. and Terrenato N. 1998. The Romanization of Italy: global accultura- to the end of the 4th century A.D.). Britannia XXXIX:161– tion or cultural bricolage? In: Forcey C, Hawthorne J, Witcher 191. R, editors. TRAC 1997 Proceedings of the Seventh Annual Redfern RC. 2010. A regional examination of surgery and frac- Theoretical Roman Archaeology Conference, Nottingham ture treatment in Iron Age and Roman Britain. Int J Osteoar- 1997. Oxford: Oxbow Books. p20–27. chaeol 20:443–471. Toller H. 1977. Roman lead coffins and ossuaria in Britain. Redfern RC, DeWitte S. 2011. A new approach to the study of Oxford: British Archaeological Reports 38. Romanization in Britain: a regional perspective of cultural Tomlin RSO. 1993. Graffiti on Roman pottery. In: Woodward PJ, change in late Iron Age and Roman Dorset using the Siler Davies SM, Graham AH, editors. Excavations at the Old and Gompertz-Makeham models of mortality. Am J Phys Methodist Chapel and Greyhound Yard, Dorchester, 1981– Anthropol 144:269–285. 1984. Dorchester: Dorset Natural History and Archaeological Redfern RC, Hamlin C, Athfield NB. 2010. Temporal changes in Society Monograph. p284. diet: a stable isotope analysis of late Iron Age and Roman Toynbee JMC. 1996. Death and burial in the Roman world. Lon- Dorset, Britain. J Archaeol Sci 37:1149–1160. don: Thames and Hudson. Redfern RC, Roberts CA. 2005. Health in Romano-British urban Ubelaker DH. 1989. Human skeletal remains: excavation, anal- communities: reflections from the cemeteries. In: Smith DN, ysis and interpretation. Washington: Taraxacum. Brickley MB, Smith W, editors. Fertile ground: papers in hon- van der Veen M, Livarda A, Hill A. 2008. New plant foods in our of Susan Limbrey. Oxford: Oxbow Books. p115–129. Roman Britain: dispersal and social access. Environ Archaeol Revell L. 2005. The Roman life course: a view from the inscrip- 13:11–36. tions. Eur J Archaeol 8:43–63. Vaupel JW, Manton KG, Stallard E. 1979. The impact of hetero- Richards MP, Molleson TL, Vogel JC, Hedges REM. 1998. Stable geneity in individual frailty on the dynamics of mortality. isotope analysis reveals variations in human diet at the Demography 16:439–454. Poundbury Camp cemetery site. J Archaeol Sci 25:1247–1252. Vaupel JW, Yashin AI. 1985. Heterogeneity’s ruses: some surpris- Saunders SR, Barrans L. 1999. What can be done about the ing effects of selection on population dynamics. Am Stat 39:176. infant category in skeletal samples? In: Hoppa RD, Fitzgerald Walker PL. 1995. Problems of preservation and sexism in sex- CM, editors. Human growth in the past: studies from bones ing: some lessions from historical collections for palaeodemog- and teeth. Cambridge: Cambridge University Press. p183– raphers. In: Saunders SR, Herring A, editors. Grave reflec- 209. tions: portraying the past through cemetary studies. Toronto: Scheidel W. 1997. Quantifying the sources of slaves in the early Canadian Scholars’ Press. p 31–48. Roman Empire. J Roman Stud 87:156–169. Wainwright GJ. 1979. Gussage All Saints: an Iron Age settle- Scheidel W. 2009. Disease and death in the ancient city of ment in Dorset. London: Department of the Environment Rome. Princeton/Stanford Working Papers in Classics. Archaeological Reports No 10.

American Journal of Physical Anthropology 208 R.C. REDFERN AND S.N. DEWITTE

Wells C. 1992. The Roman Empire. London: Fontana Press. Wood JW, Milner GR, Harpending HC, Weiss KM. 1992. The Wiedemann T. 1989. Adults and children in the Roman Empire. osteological paradox: problems of inferring prehistoric health London: Routledge. from skeletal samples. Curr Anthropol 33:343–370. Wilkinson R, Marmot M. 2003. Social determinants of health: Woodward PJ, Davies SM, Graham AH, editors. 1993. Excava- the solid facts. Copenhagen: World Health Organization. tions at the Old Methodist Chapel and Greyhound Yard, Dor- Wood JW, Holman DJ, O’Connor KA, Ferrell RJ. 2002. Mortal- chester, 1981–1984. Dorchester: Dorset Natural History and ity models for paleodemography. In: Hoppa RD, Vaupel JW, Archaeological Society Monograph Series. editors. Paleodemography: age distributions from skeletal Woolf G. 2005. Family history in the Roman North-West. In: samples. Cambridge: Cambridge University Press. p 129– George M, editor. The Roman family in the Empire: Rome, 168. Italy, and beyond. Oxford: Oxford University Press. p 231–254.

American Journal of Physical Anthropology